[R-sig-ME] Need some help with glmer output (Hans Ekbrand)

[Hans Ekbrand]

On Sun, Feb 19, 2012 at 09:47:29PM -0800, lopez toledo wrote:
> Thanks Hans for your two messages to my question on glmer yesterday. May I ask another question?

Sure, I was just thinking about my second suggestion, it required the
Time variable to have certain properties that you did not explictly
state it had.

> In your first reply you suggested to include only one single random effects (Palm) and Time as Fixed factor (1). I like that model as I can see the effects among years.  However in your second suggested model, Time as random effect does not indicate to me whether there is difference among years. There is some variance, but variance does not say nothing to me, specially if there is not a total/residual variance to compare with!

If you like the first model, then use it. The second version only makes sense
if there are idiosyncracies of the years, e.g. very little rain in
year 2010 or something like that.

If you're only interested in general differences in growth rate
between plants that are 1, 2, and 3 years old, then the first model is
what you want.

> 1) Model1<-glmer(Total leaves ~ DT * G * Time + (1 | palm), family=poisson)
> 2) Model1<-glmer(Total leaves ~ DT * G + (1 | Time) + (1 | palm), family=poisson)
> 
> Can you let me know your thought?
> 
> Leo